Serious Games for Museum Environments Anastasios Doulamis1, Fotis Liarokapis2, Panagiotis Petridis3, Georgios Miaoulis4
1: Technical University of Crete, Decision Support Lab. Chania, 73100, Crete, Greece Tel: + 30 28210 37430 Email:
[email protected] 2: Coventry University Coventry, United Kingdom, UK Interactive Worlds Applied Research Group Tel: +44 (0)24 7688 7631 Email:
[email protected] 3: Serious Games Institute (SGI) Coventry Innovation Village Coventry University Technology Park Cheetah Road, Coventry, West Midlands, CV1 2TL Coventry, United Kingdom, UK Email:
[email protected] 4: Technological Education Institute of Athens Department of Computer Science Ag.Spyridonos St., 122 10 Egaleo, Greece Tel: (+30) 2 10 53 85 312, Fax: (+30) 2 10 59 10 975 Email:
[email protected]
Abstract: This paper presents a serious game for cultural heritage and in particular for museum environments focusing on the younger generation. The aim of the game is to solve a treasure hunt scenario by collecting medieval objects that used to be located in and around the Priory Undercroft. Located in the heart of Coventry, UK, the Priory Undercrofts are the remains of Coventry’s original Benedictine monastery, dissolved by Henry VIII. Initial user testing demonstrated the potentials of serious games for education in museum environments. Game has been created using innovative aspects as far as 3D content reconstruction is concerned as well as educational scenarios. Thus, the main contribution of this paper lies in the direction of accelerating 3D reconstruction by combining computer vision tools and on the educational dimension of the developed game. In addition, artificial intelligence tools are adopted for conducting the plot of the game. Keywords: serious games, computer graphics, 3D modelling, artificial intelligence, cultural heritage
1. INTRODUCTION It is well known that we live in a threedimensional (3D) world and we perceive most of the events/activities/actions via exploiting depth information. Among all human senses, the visual, and especially the 3D one, is probably the most important in precisely comprehending our world. This abstractive and hierarchical way of thinking of humans is due to “the modularity of mind”, as is supported in the pioneering work of Dr. Fodor [1]. According to this theory, our brain follows a hierarchical way of thinking; starting from the simplest thought and ending at the most advanced one. Specifically, this theory implies that if, for example, we want to describe a red rose in a bunch of flowers, the simplest “concept” coming to our mind is . . . an image! To be more specific, it is a 3D image, which clearly consists of all the information we want to describe and which is derived from the experience of seeing a red rose. In other words, a 3D object contains all the ground truth of a physical object, while its 2D views provide only a subset of the 3D object, an abstraction of the real world [2]. 3D models have nowadays become ubiquitous for applications such as computer games, robotics, machine vision, computeraided design, cultural heritage, architectural design and planning [3]-[8]. We live in an era where the acquisition of 3D data is ubiquitous, continuous, and massive. These data come from multiple sources including highresolution, geo-corrected imagery from aerial photography and satellites. Now researchers are developing mobile geo-located systemssome containing calibrated laser range finders and cameras-that can collect ground-level detail at unprecedented resolution. In the near future, individuals or robots will be equipped with stereoscopic cameras and other sensors as well as the computational power to pervasively collect and organize geolocated data [9].
3D world has generally stimulated different types of applications. Some are related with 3D reconstruction, either precise or approximate and some with computer games technologies. Precise 3D reconstruction (that should resemble as much as possible the real 3D objects) usually demands high degree of humans’ interaction that makes 3D computer vision tools be of high cost (manual effort) [10]. Thus it demands huge computational load. For this reason, despite the recent research efforts, 3D computer vision remains one of the most arduous problems in the computer engineering society, especially for real-world applications in which we cannot impose constraints on lighting conditions, object occlusions, and possible dynamic modifications of the environment. This is, for instance, the case of the Future Internet infrastructure which is now moving towards a 3D media cyberspace, 3D content-based visual retrieval and indexing applications, and processes for automatic understanding events and actions in 3D video streams as close as possible to humans’ perception [11]. In particular, (i) Future Internet is moving to a 3D media internet. As the Internet has revolutionized the access to multimedia content and enabled collaborative usergenerated content, it is now evolving towards a 3D infrastructure. 3D processing will enable mass distribution and caching of 3D content and enhanced user quality of experience with optimized impact on the performance of the underlying processing and networking platforms. (ii) Despite the current photogrammetric tools for 3D reconstruction, which are tools of high computational cost, 3D media internet needs new “easy and affordable components” for representing the 3D content, algorithms applicable to low-cost devices (of relatively low capabilities) and/or manageable by non expert users. (iii) In the same framework, the recent advances in geographic information systems and in multimedia technologies have made exigent the need for reconstructing 3D data for an
extremely vast volume of data (e.g., entire towns). Today, such a 3D digitalization is practically forbidden due to the high computational cost and effort required. On the other hand, computer games with complex virtual worlds for entertainment have been widespread today, especially amongst the younger generation. These advances have been verified by both the recent achievements in software and hardware technologies. These led to a rise in the quality of real-time computer graphics and increased realism and immersion in computer games. One particular interesting application of 3D content modelling refers to cultural heritage reconstructions. It is the preservation of historical sites and monuments from erosion, vandalism, and other long-lived artifacts that usually cause damages and need for repair. It’s important to keep an accurate record of these sites’ current conditions by using 3D model building technology, so preservationists can track changes and/or predict structural problems [10]. The main problem in reconstructing an accurate 3D model is that this process is tedious involving a lot of manual effort and thus making the problem for 3D reconstructing of huge areas (like entire cities) a real-life demanding problem. Similarly, although the widespread use of gaming for leisure purposes has been well documented, the use of games to support cultural heritage purposes, such as historical teaching and learning, or for enhancing museum visits, has been less considered [12]. As a result, it is important to introduce the concept of serious games to support cultural heritage purposes. The potential use of serious games technologies on cultural heritage education has been recently addressed in the literature and particularly in papers [13][14][15]. A reason for that is the popularity of video games among the younger generation ranging between 10 to 30 years old. Thus, they would
be an ideal means for educational purposes. The term ‘serious games‘ describe exactly this concept. The use of a pedagogical game that can be use to teach people under an informal learning scenario. Typical examples are game engines and online virtual environments that have been used to design and implement games for non-leisure purposes, e.g. in military and health training [16][17]. In this paper, we present a new framework for serious games applications in cultural heritage. The developed algorithms allows for real-time interactive visualization and simulation of realistic virtual heritage scenarios, such as 3D reconstructions. Examples include ancient sites and monuments, using off-the-self-components. The serious game is finally evaluated using a small sample of students. Before describing the goals and contribution of this paper, we first cite the most representative works reported in the literature in this theme presenting state-of-the-art papers.
2. PREVIOUS WORKS Both, entertainment games technologies and serious games technologies share common state-of-art concepts. As the work of [16] mentions, the only difference between serious games techniques and entertainment game structures is the application of games technologies to a non-entertainment domain. However, the difference can be generalized as contributions in the areas of visual expressions, communications and collaboration mechanisms should be properly and promptly examined. Recent advances on gaming technology are tremendous. The contemporary graphics can achieve in real-time (or at least in a just-in-time framework) near optimal photorealism. This leads to a dramatic increase of virtual games worlds populated with rich multimedia content that considerable improves the quality of experience for the users. And what is common in serious games developers is that it is important to strengthen entertainment, fun and pedagogy [18]. In other words, there is a need for the game developers
and instructional designers to work together to develop engaging and motivating serious games for the future.
Figure 1: A snapshot of the Rome Reborn project 1
Moreover, 3D reconstruction for historical sites is an approach that has been studied previously [19][20]. Several funded projects have been developed towards this direction. However, these systems still remain within the academic community without being released to enterprises and/or to wide public for a commercial exploitation. One of the largest 3D reconstruction projects is the one dealing with the reborn of Ancient Rome. The main aims of the project are to produce a high resolution version of Rome at 320 AD (Figure 1), a lower resolution model for creating a ’mashup‘ application with ’Google Earth‘ (http://earth.google.com/rome/), and finally the collaborative mode of the model for use within virtual environments and aimed primarily at education [21]. Towards the same field, the ancient Pompeii (a Roman city, which was destroyed and completely buried in the first recorded eruption of the volcano Mount Vesuvius in 79 AD) was reconstructed [20]. The main goal of this project was to simulate a crowd of virtual Romans exhibiting realistic behaviours in a reconstructed district of Pompeii. Similarly, the Parthenon project reconstructs the Minerva’s Temple in the Athenian Acropolis creating a virtual version of the Parthenon and its separated sculptural elements.
1
http://www.romereborn.virginia.edu/gallerycurrent.php
Other types of applications are dealing with the creation of virtual museums using computer games technologies. A recent survey paper that examines all the technologies and tools used in museums was recently published [24]. Here we present several examples of this type of cultural heritage serious game, including some virtual museums that can be visited in real-world museums. A characteristics example includes the game of [25] depicts a hypothetical Virtual Egyptian Temple (no real-world equivalent) embodying all of the key features of a typical New Kingdom period Egyptian temple in a manner that an untrained audience can understand. The game provides enough pedagogical questions that ameliorates the quality of experience for the player and simultaneously increase his/her educational background. In the same framework, the Foundation of the Hellenic World has produced a number of games related with the Ancient Olympic Games [26]. The technology needed to produce all these historical games are based on a game engine which provides the core technology for the creation and control of the virtual world. A game engine is an open, extendable software system on which a computer game or a similar application can be built. It provides the generic infrastructure for game creation [16], i.e. I/O (input/output) and resource/asset management facilities. The possible components of game engines include, but are not limited to: rendering engine, audio engine, physics engine, animation engine. 3D rendering is another important aspect of a serious game, similar to the entertainment games; it requires a lot of graphical features and effects. The state-ofthe-art in this subject area is broad and, at times, it can be difficult to specify exactly where the ‘cutting edge’ of the development of special effects lies. A number of the techniques that are currently in use were originally developed for offline applications and have only recently become adopted for
use in real-time simulations through improvements in efficiency or hardware.
•
The Content Management Module
•
The Artificial Intelligent Module
3. GAME DESCRIPTION Built on the site of the Benedictine Priory of St Mary’s the Centre tells the story of Coventry’s first Cathedral, founded in the 11th century by Lady Godiva and Earl Leofric. The centre tells the story of using archaeological finds discovered during the excavation of the site. Located in the heart of Coventry, UK, the Priory Undercrofts are the remains of Coventry’s original Benedictine monastery, dissolved by Henry VIII. The Priory Visitor Centre, designed by architects MacCormac, Jamieson and Prichard, brings together objects, images and the latest information about St Mary's Benedictine Cathderal and Priory. Inside the Visitor Centre hangs a mobile of coloured glass by John Reyntiens. It was made using some of the same techniques as the makers of medieval stained glass windows and incorporates the shapes of angel eyes and wings as inspired by the wall painting from the Chapter House. The Priory Undercrofts include remains of the original vaulting, windows and a fireplace. The motivation is to raise the interest of the younger generation in the museum, as well as cultural heritage in general. The aim of the serious game is to solve a treasure hunt scenario by collecting medieval objects that used to be located in and around the Priory Undercroft. Each time a new object is found, the player is prompted to answer a question related to the history of the site. A typical user-interaction might take the form of: “What did St. George slay? – Hint: It is a mythical creature. – Answer: The Dragon”, meaning that the user then has to find the Dragon. The system architecture, which is presented in Figure 2, is composed by three modules including: •
The Visualization Module
Figure 2: System Architecture
The Priory Undercroft Visualization module is based on the Quest3D visualization engine. Quest3D is a very flexible authoring environment for real-time 3D applications. The edit-while-executing and graphical nature of Quest3D makes it one of the most intuitive tools to work with. Quest3D is used by developers, educational institutions and VR companies [27]. The content management system (CMS) is implemented within the authoring tool of QUEST3D and allows the educator to create various learning materials. The CMS is also responsible for the organization of tasks that the user will complete during a particular scenario. Tasks are organized according to a pre-planned decision tree (Figure 3). Each scenario step may contain one or more decisions sets. Each decision set is connected to the next scenario step by a transition that occurs only when all decisions of a given decision set has been taken. In some cases, a decision set may contain a single decision. In this case, the decision leads to the next scenario step.
solve in order to progress to the next puzzle and finish the games.
Figure 3: Decision Tree [28]
Puzzle
Typical Puzzle Question
1
Find the statue of Saint George
2
What did Saint George Slay? Hint: It is a mythical Creature
3
What does a dragon breath
4
Who might have warmed by the fire
5
Monks had had important jobs that took time and precision. What did they create
6
To create perfect illuminated texts the monks had to write in straight lines. They did not use rulers. Can you find the tool they used?
7
What other form of decoration might you find where the monks gathered
8
To show who created a piece of artwork or craft the artist might do what?
Table 1 Different Puzzles which the player has to solve
A decision tree strikes the balance between “user flexibility” and “writer flexibility” [29]. User flexibility, refers to the learner’s range of possible actions, in order to make a real-time environment as realistic as possible, this range of action should be maximised. Writer flexibility refers to the developers control over the scenario, ensuring that intended learning outcomes are presented during the scenario. The learner is guided through the scenario with a series of puzzles which they have to
Table 1 gives an example of the puzzles that the learner has to solve to progress to the next one. Learner login information is also stored within the database, which will form the basis of a learner tracking system. This tracking system will record decisions and actions made by the user throughout a simulator session for later interrogation by the educator. The exhibition side is based on the Bergeron [30] principles for a good game interface. Following the guidelines of Bergeron, the exhibition side of the Priory Undercroft was created with the user in mind. The Priory Undercroft graphical interface is illustrated on Figure 4.
Figure 4: Priory Undercroft Game in Operation
The Artificial Intelligence module is based on the 3 Level of Interaction Framework (LoI) which was developed in collaboration between the Serious Games Institute and Toulouse University [31]. The LoI framework simplifies the interaction between the player and the non player characters (NPC). Graphically, the LoI can be represented as auras of increasing complexity centered on the player’s avatar (see Figure 5). LoI is based on a simple social space metric [31] and is divided to three levels. The first level aims to populate the characters with authentic crowd in order to increase the immersion of the player. Characters located in closer surrounding of the player belong to the
interaction level. Finally, a character inside the dialogue level interacts with the player in a natural way, ultimately using speech recognition and synthesis. All the NPC by default belong to the background level, but as the player moves on the environment and they happen to get closer or away from the player and thus enter or exit the interaction or dialogue levels.
Figure 5: Level of Interaction Framework
4. 3D RECONSTRUCTION Many cultural heritage applications require 3D reconstruction of real-world objects and scenes. This is also the case reconstructing a historical site as in this paper. The recent advances in 3D digitalization and modelling and the recent technological evolutions in laser-scanning, 3D modelling software, imagebased modelling techniques, computer power, and virtual reality has permitted the detailed and accurate 3D reconstruction of such places. Many approaches are currently available, mainly exploiting CAD tools and/or traditional photogrammetry with control points and a human operator. However, this approach is time-consuming and can be costly and impractical for large-scale sites. Modelling methods based on laser-scanned data and more automated image-based techniques have recently become available [32]. In our approach, we efficiently combine the traditional laser-scanning approaches with computer vision methods. The latter method is based on the finding of relations among similar geometric patterns in an image. That is, we find the correspondences of characteristic
(salient) points between two images that share a part of common content so as to accelerate the time in the reconstruction process. To acquire data describing an entire structure such as historical site requires taking multiple range scans from different locations that we must register together correctly. Although we can register the point clouds manually, this process is a time consuming and error prone [10]. That is, manually visualization of vast amount of points will surely result in erroneous situations. Initially, for each image a set of salient points are detected by applying on the visual content scale invariant transformations. Characteristic examples are the Scale-Invariant Feature Transforms (SIFT) which select for any object in an image. Interesting points on the object can be extracted to provide a "feature description" of the object [33]. This description, extracted from a training image, can then be used to identify the object when attempting to locate the object in a test image containing many other objects. The most important issue in SIFT is the fact that the extracted features are invariant under scale, illumination and/or noise alterations. We also extract the Histograms of Oriented Gradients (HOGs). This technique counts occurrences of gradient orientation in localized portions of an image. This method is similar to that of edge orientation histograms, scale-invariant feature transform descriptors, and shape contexts, but differs in that it on a dense grid of uniformly spaced cells and uses overlapping local contrast normalization for improved accuracy [34]. Let us then denote as W , V ∈ R 2 two sets of two-dimensional (2D) points of the aforementioned features detecting onto two images that share some common parts. Assuming without loss of generality that the two sets include the same number of points and assuming that there exist a known
mapping, say A that relates one by one each feature point of set W with each point in set V, then, in statistical theory we can prove that we can estimate the optimal geometric alignment between the two sets.
G (W , V ; A) = max ∑ A(B ⋅ w, A(w )) (1) B w∈W
where w ∈ W is a 2D point in set W expressed in homogeneous coordinates. Similarly, B is a transformation matrix. We express the content in homogeneous coordinates since in this space we can get as a matrices product all the affine transformations (scaling, rotation, skewing and translation) including vertical and/or horizontal translation. On the contrary, we use of conventional Cartesian coordinates violates the product property for the simple translation transformation. The main problem of (1) is that it never appears as such in our case, due to unknown feature correspondence and outliers [35]. This problem can be overcome by calculating the correspondences among the two sets W , V ∈ R 2 , as we do in finding the disparity field map. In particular, if we denote as v ∈ V a 2D point in set V and we recall that w ∈ W is a 2D point in set W both expressed in homogeneous coordinates. Then, the correspondences are found by estimating the points in V that best matches with the points in W. In other words,
G (W , V ; A) = max ∑
∑ cv, wd (v, w) (2)
c v∈V w∈W
In (2) d ( v, w ) is an arbitrary similarity measure. This is a very important problem because it can work well in practice if the features are discriminative enough. It can be proved in [36] that equation (2) can be expressed as an inner product of two independent histograms the contain information about the elements of the two sets. In case that the histograms are
normalized this coincides with the cosine similarity measure [37]. To improve the correspondences among the two point sets, we developed in this paper an extension of the RANdom SAmple Consensus (RANSAC) algorithm [38] which yields fast performance relying on the approach of [39]. The work initially refines normals of the point clouds, even in the presence of noise. Then, it computes scores on local planes for each point. Selecting the best local planes and applying a region growing approach we can estimate the correspondences among the different clouds.
5. GAME EVALUATION To acquire feedback on the finished core of the application, a self-contained executable file was supplied to a small number of Coventry University students based on the hallway usability testing methodology. The intention of these tests was primarily to gather information on the playability and enjoyability of the game, but also to discover potential technical problems. All of the endusers had some experience with games, and the vast majority described themselves as ‘gamers’. A few of those involved also had experience with games programming, or had some knowledge of the architecture behind creating a game. For all users, the aim of the game was presented and it was explained that the players should not expect a complete game, but rather a prototype. Five students from the Faculty of Engineering and Computing, Coventry University were asked to participate in the test group. The average time of the tests was approximately 30 minutes. Instead of asking the University students specific questions, they were asked to talk through what they were doing and how they felt as they played the game. Overall, recorded feedback was very encouraging and all users agreed that the serious game has a lot of potential for cultural heritage applications. They also mentioned
that they prefer the idea of ‘playing’ and ‘learning’ at the same time. On the other hand, a number of important issues were pointed out. In particular, one student had some minor issues with the controls, especially the combination of the mouse and keyboard to navigate inside the environment. It is worth-mentioning that after playing the serious game for long enough, the player adjusted to the issue without any further problems. Another user commented that he would play such a game, on the condition that further additions were made to the game play. On the positive side, all users agreed the educational aspect of the game is obvious and helps them to understand and learn something about the history of Priory Undercrofts.
6. CONCLUSIONS AND FUTURE WORK This paper presented the architecture of a serious 3D game for museum environments focusing on the younger generation. The aim of the game is to solve a treasure hunt scenario by collecting medieval objects that used to be located in and around the Priory Undercroft. Initial user testing demonstrated the potentials of serious games for education in museum environments. In the future we will add more intelligence to the game, based on intelligent avatars. An online version will be also developed so that it can be accessed remotely. Finally, the game will be installed inside the Herbert Art Gallery & Museum so that it can be evaluated by the museum visitors.
ACKNOWLEDGEMENTS The authors would like to thank the following: The Herbert Art Gallery & Museum (Coventry, UK), Simon Bakkevig, and Lukasz Bogaj.
REFERENCES [1] J. A. Fodor, The modularity of mind. Cambridge: Cambridge Bradford Books, MIT Press. [2] P. Daras, A. Axenopoulos, “A 3D Shape Retrieval Framework Supporting Multimodal Queries”, International Journal of Computer Vision, Springer Press, DOI 10.1007/s11263-009-0277-2, Jul 2009. [3] Real-time 3D models, http://www.3drt.com/. [4] S. Jayanti, K. Kalyanaraman, N. Iyer, and K. Ramani, “Developing an engineering shape benchmark for CAD models,” Computer-Aided Design, 38(9), 939–953. [5] N. D. Doulamis, G. Bardis, J. Dragonas, G. Miaoulis, D. Plemenos, “Collaborative Evaluation Using Multiple Clusters in a Declarative Design Environment,” Artificial Intelligence Techniques for Computer Graphics 2008: 141-157 [6] N. D. Doulamis, G. Bardis, J. Dragonas, G. Miaoulis, “Optimal Récursive Designers' Profile Estimation in Collaborative Declarative Environment,” IEEE International Conferences on Tools of Artificial Intelligence, Vol. 2, pp. 424-427, 2007. [7] V. Douskos, L. Grammatikopoulos, I. Kalisperakis, G. Karras, E. Petsa, “FAUCCAL: An Open Source Toolbox for Fully Automatic Camera Calibration,” XXII CIPA Symposium on Digital Documentation, Interpretation & Presentation of Cultural Heritage, October 11-15, Kyoto, Japan, 2009. [8] V. Douskos, I. Kalisperakis, G. Karras, E. Petsa, “Fully automatic camera calibration using regular planar patterns,” International Archives of Photogrammetry, Remote Sesnisng and the Spatial Information Sciences, Vol. 37, No. 5, pp. 21-26, 2007 [9] B. Ribarsky, H. Rushmeier, “3D Reconstruction and Visualization,” IEEE Computer Graphics and
Applications, pp. 20-21, November-December 2003. [10] P. K. Allen, A. Troccoli, B. Smith, S. Murray, I. Stamos and M.s Leordeanu, “New Methods for Digital Modeling of Historic Sites,” IEEE Computer Graphics and Applications, pp. 32-41, NovemberDecember 2003. [11]
Cordis.europa.eu//workpackage
[12] E. F. Anderson, L. McLoughlin, F. Liarokapis, C. Peters, P. Petridis, S. de Freitas, “Serious Games in Cultural Heritage,” The 10th International Symposium on Virtual Reality, Archaeology and Cultural Heritage VAST, 2009. [13] T.H. Apperley, “Virtual unaustralia: Videogames and australias colonial history,” in Proc. Of Cultural Studies Association of Australasias Annual (UNAUSTRALIA), 2006. [14] R. Francis, “Revolution: Learning about history through situated role play in a virtual environment,” In Proceedings of the American Educational Research Association Conference, 2006. [15] J. Jacobson, L. Holden, “The virtual egyptian temple,” In ED-MEDIA: Proccedings of the World Conference on Educational Media, Hypermedia & Telecommunications, 2005. [16] M. Zyda, “From visual simulation to virtual reality to games,” IEEE Computer, Vol. 38, No. pp. 25-32, 2005. [17] M. Macedonia, “Games Soldiers Play,” IEEE Spectrum, Vol. 39, No. 3, pp. 32-37, 2002. [18] S. de Freitas, and M. Oliver, “How can exploratory learning with games and simulations within the curriculum be most effectively evaluated?” Computers and Education, Vol. 46, pp. 249-264, 2006 [19] D. Arnold, A. Day, J. Glauert, S. Haegler, V. Jennings, B. Kevelham, R. Laycock, N. Magnenat, N. Thalamnn, J. Maim, D. Maudu, G, Papafiannakis, D.
Thalmann, B. Yersin, K. Rodriguez-Echavarria, “Tools for populating cultural heritage environments with interactive virtual humans,” In Open Digital Cultural Heritage Systems, EPOCH Final Event Rome (2008). [20] J. Maim, S. Haegler, B. Yersin, P. Mueller, D. Thalmann, L. van Gool, “Populating ancient Pompeii, with crowds of virtual Romans,: In VAST07: The 8th International Symposium on Virtual Reality, Archaeology and Intelligent Cultural Heritage, pp. 109–116, 2007. [21] B. Frischer, “The Rome reborn project. How technology is helping us to study history,” OpEd, November 10, University of Virginia, 2008. [22] P. Debevec, “Making "The Parthenon". 6th International Symposium on Virtual Reality, Archaeology, and Cultural Heritage, 2005 [23] G. Jones, M. Christal, “The future of virtual museums,” On-line, immersive, 3D environments. Created Realities Group, 2002. [24] S. Sylaiou, F. Liarokapis, K. Kotsakis, and K. Patias, “Virtual museums, a survey on methods and tools,” Journal of Cultural Heritage, 2009. [25] J. Jacobson, L. Holden, “The virtual Egyptian temple,” Proccedings of the World Conference on Educational Media, Hypermedia & Telecommunications (ED-MEDIA), 2005. [26] A. Gaitatzes, D. Christopoulos, G. papaioannou, “The Ancient Olympic Games: Being Part of the Experience.,” The 5th International Symposium on Virtual Reality, Archaeology and Cultural Heritage (VAST) pp. 19–28. 2004. [27] Quest3D. 2009 [06/11/2009]; Available from: www.quest3d.com. [28] Ponder, M., Herbelin, B., Molet, T., Schertenlieb, S., Ulicny, B., Papagiannakis, G., Magnenat-Thalmann, N., and Thalmann, D. Immersive VR decision training: telling interactive stories featuring advanced virtual human simulation technologies. in EGVE '03: Proceedings
of the workshop on Virtual environments. 2003. Zurich, Switzerland, May 22 - 23, 2003: ACM. [29] Magerko, B., Laird, J., Towards Building an Interactive, Scenario-based Training Simulator. in 10th Computer Generated Forces and Behavior Representation Conference. 2002. Orlando, FL. [30] Bergeron, B.P., Developing Serious Games. Game Development Series. 2006, Massachucetts: Charles River Media, Inc. [31] Panzoli, D., et al., A Level of Interaction Framework for Exploratory Learning with Characters in Virtual Environments, in Intelligent Computer Graphics 2010, D. Plemenos and G. Miaoulis, Editors. 2010, Springer Berlin / Heidelberg. p. 123-143. [32] F. Sabry, El-Hakim, J.-Angelo Beraldin, M. Picard, and G. Godin, “Large-Scale Heritage Sites with Integrated Techniques,” IEEE Computer Graphics and Applications, pp. 21-29, May-June 2004. [33] L. David, “Object recognition from local scale-invariant features,” Proceedings of the International Conference on Computer Vision, 1999.
[34] N. Dalal, B. Triggs, “Histograms of oriented gradients for human detection,” IEEE Computer Vision and Pattern Recognition Conference, CVPR, 2005. [35] I. L. Dryden and K. V. Mardia, “Statistical Shape Analysis,” Wiley Blackwell, July 1998. [36] H. Jegou, M. Douze, and C. Schmid, “Improving bag-of-features for large scale image search,” International Journal of Computer Vision, pp. 1-21, 2010. [37] Shiwei Zhu, Junjie Wu, Guoping Xia, “TOPK cosine similarity interesting pairs search,” 2010 Seventh International Conference on Fuzzy Systems and Knowledge Discovery (FSKD), Beijing, China, 2010. [38] A. David Forsyth and J. Ponce. Computer Vision, a modern approach. Prentice Hall. ISBN ISBN 0-13-085198-1, 2003. [39] J.-E. Deschaud, “A Fast and Accurate Plane Detection Algorithm for Large Noisy Point Clouds Using Filtered Normals and Voxel Growing,” 5th International Symposium on 3D Data Processing, Visualization and Transmission Espace SaintMartin, Paris, France, May 17 - 20, 2010.